Note that one of the first responders blamed the blockage for delaying care and potentially contributing to the death: 'Collectively, these interferences “contributed to a poor patient outcome, delaying the definitive care required in severe trauma cases,” according to one of the reports.'
Yeah how many have they saved to date? It’s a valid point and I’d love to know the answer now that we know self-driving vehicles are responsible for >= 1 person’s death
I don't know how you'd even measure the numbers of deaths that would've resulted from accidents that didn't happen because an autonomous car avoided it in a situation where an inattentive, reckless, or unskilled human driver wouldn't have... especially without an otherwise identical control environment.
Or even situations where partial self-driving (such as Tesla's system) reduced driver fatigue that would've later contributed to a more exhausted driver causing an accident.
I'm the furthest thing from a Tesla fanboy, but I did try one for a while in order to have a more unbiased view and I will say that "autopilot" subjectively reduced my stress and fatigue during and after 5 mph bumper to bumper traffic jams and left me feeling more attentive afterwards compared to those same types of traffic conditions in cars being operated 100% by me.
I'd assume that a large part of the data that makes up the "mile" portion of the passenger per mile is large trips. So, look at traffic deaths in some dense cities and you will probably find much larger numbers than you'd expect.
As a quick check of this idea, looking at "vehicle passenger deaths atlanta" in a search gives numbers around 100 for many years. (And Atlanta isn't that dense, all told.)
Being able to handle 5 mph bumper to bumper traffic jams is hardly a Tesla-specific feature. It's quickly becoming a standard feature in cars, and such driver assistance features are rather independent from the conversation about full self-driving (whether via Cruise cars or Tesla FSD).
In normal circumstances, this would be easy. Just look at the base rate of deaths and see how the needle moved after these were introduced. This is how we know, for example, that automatic emergency brakes are saving lives.
Unfortunately, the pandemic through normal out the window. :( Probably still possible to find some natural experiments where you can control some variables. My gut would be that their are some comparable cities that differ mainly in whether they have these cars. That said, it will be a lot harder, as the impact of the pandemic was pretty massive.
This article does not have enough information about the base rate. It is impossible to assess this situation without knowing how often do human drivers get in the way of emergency service vehicles.
well...yes? For example: wouldn't you be more upset if a collision was caused by a drunk driver than by a driver who was paying attention? That's because drunk drivers are known to be dangerous and should not be operating.
If self-driving vehicles overall increase safety, then I am not as upset as I am if they are known to overall decrease safety, because the fact of their presence being just or unjust is pertinent to the case.
The question is whether it should come up at all. If my ambulance takes a while to nudge its way through some crowded traffic at a busy intersection, would anyone think to tell my family that poor urban design and a lack of roundabouts contributed to my death?
While parked or double parked cars can get in the way of emergency service vehicles (that seems to have been the case here with an unoccupied police car contributing to the blockage together with the two Cruise cars), in most cases human drivers will be in the car and can generally be cleared out with the use of a siren.
As to Cruise cars themselves, it's been happening often enough that SF emergency services have been repeatedly complaining about it.
In defense of Cruise though, the pedestrian that died here was struck by a human-driven vehicle, and presumably such pedestrian strikes would be a lot less common if human drivers were replaced by Cruise cars.
That's simply not a defense of Cruise. How things work in a hypothetical universe doesn't absolve them of how things worked in this universe. It's a self-driving car. One of the things it should be able to do is move out of the way.
The conversation we should be having is how to give emergency responders the ability to move the car on their own. From a technical stand point, that's no problem at all. From a security stand point, it's a moderately difficult problem. From a civil rights stand point, it's a very difficult problem.
> in most cases human drivers will be in the car and can generally be cleared out with the use of a siren.
vs.
> Two autonomous Cruise vehicles and an empty San Francisco police vehicle were blocking the only exits from the scene
from the article (ie, humans at 50% of the danger of autonomous vehicles, having selected from an article that may be a hit piece on autonomous vehicles). The base rate for human-driven vehicles blocking ingresses and egresses is probably quite high.
I've seen ambulances with sirens blaring stuck in traffic. It just can't be that uncommon. Even before worrying about what madness the low-end of human drivers are capable of.
> I've seen ambulances with sirens blaring stuck in traffic
Yet another reason to accelerate the transition to self driving cars. The weirdness around emergency vehicles will be transitory. Once we hit critical mass, ambulances will basically never have issues like that again
That’s not a relevant comparison.
The article is about Cruise cars that stop dead in their tracks with nobody able to physically move the car and blocking an exit path (several minutes +} You’re bringing up a comparison of someone who may delay the responders from getting to their destination by seconds.
A human in a car can and will often move within seconds. Autonomous cars are stopping dead in their tracks. A first responder should be able to take control of that vehicle and move it. But they can’t. So they’re stuck.
> A human in a car can and will often move within seconds.
Literally the situation in the article has a human-operated car that was blocking the ambulance. There were 3 cars, 2x automated and one human-controlled.
I suspect the human-controlled one did not move in seconds or they wouldn't be talking about delays being a factor in this tragic loss of life. The fact that you didn't pick that up is what makes me think the article is a hit piece.
I hope Mayor Breed and the San Francisco DA hold the family of the deceased pedestrian accountable for their loved one's actions, and for any damage this negative publicity does to the reputation of Cruise self-driving cars.
>4. SFFD members had to locate an SFPD officer and request him to move his vehicle to allow successful egress from the scene, but doing so further delayed patient care.
Sorry, a human driver fatally hit a pedestrian, then a SFPD car was blocking the ambulance from leaving with the injured person? and we're going to make this story about the dangers of Cruise cars? I'm all for complaining if the cruise car did something wrong, but vehicles get disabled in the road all the time for various reasons.
> Two autonomous Cruise vehicles and an empty San Francisco police vehicle were blocking the only exits from the scene, (...), forcing the ambulance to wait while first responders attempted to manually move the Cruise vehicles or locate an officer who could move the police car.
aka the SFPD car was probably there, and cruise cars stalled, blocking the exit so they had to find someone to move one of the 3 cars, 2 being cruise.
I don't think that SFPD should need to anticipate in their emergency response that 2 lanes would be blocked by 2 different Cruise vehicles. Do you also take issue with the Fire Department blocking 2 lanes to shield the patients from oncoming traffic?
The deflection is just weird to me. Can we only discuss issues with self-driving cars once they've replaced human drivers fully?
42 comments
[ 3.6 ms ] story [ 95.9 ms ] threadOr even situations where partial self-driving (such as Tesla's system) reduced driver fatigue that would've later contributed to a more exhausted driver causing an accident.
I'm the furthest thing from a Tesla fanboy, but I did try one for a while in order to have a more unbiased view and I will say that "autopilot" subjectively reduced my stress and fatigue during and after 5 mph bumper to bumper traffic jams and left me feeling more attentive afterwards compared to those same types of traffic conditions in cars being operated 100% by me.
Consequently, the amount of testing required to prove that the fatality rate of autonomous vehicles is lower than human-driven one’s is astronomical.
Non-fatal accidents are more common but the statistics are far more rubbery.
As a quick check of this idea, looking at "vehicle passenger deaths atlanta" in a search gives numbers around 100 for many years. (And Atlanta isn't that dense, all told.)
Unfortunately, the pandemic through normal out the window. :( Probably still possible to find some natural experiments where you can control some variables. My gut would be that their are some comparable cities that differ mainly in whether they have these cars. That said, it will be a lot harder, as the impact of the pandemic was pretty massive.
If self-driving vehicles overall increase safety, then I am not as upset as I am if they are known to overall decrease safety, because the fact of their presence being just or unjust is pertinent to the case.
Car infrastructure is notorious for focusing on the safety of drivers rather than pedestrians.
If self-driving cars increase safety for drivers and are worse for everybody else I am not sure it is a good trade-off.
In this regard we can adapt the saying about electric cars to self-driving cars: they are here to save the car industry not you.
As to Cruise cars themselves, it's been happening often enough that SF emergency services have been repeatedly complaining about it.
In defense of Cruise though, the pedestrian that died here was struck by a human-driven vehicle, and presumably such pedestrian strikes would be a lot less common if human drivers were replaced by Cruise cars.
The conversation we should be having is how to give emergency responders the ability to move the car on their own. From a technical stand point, that's no problem at all. From a security stand point, it's a moderately difficult problem. From a civil rights stand point, it's a very difficult problem.
vs.
> Two autonomous Cruise vehicles and an empty San Francisco police vehicle were blocking the only exits from the scene
from the article (ie, humans at 50% of the danger of autonomous vehicles, having selected from an article that may be a hit piece on autonomous vehicles). The base rate for human-driven vehicles blocking ingresses and egresses is probably quite high.
I've seen ambulances with sirens blaring stuck in traffic. It just can't be that uncommon. Even before worrying about what madness the low-end of human drivers are capable of.
Yet another reason to accelerate the transition to self driving cars. The weirdness around emergency vehicles will be transitory. Once we hit critical mass, ambulances will basically never have issues like that again
That's something that should be addressed.
Literally the situation in the article has a human-operated car that was blocking the ambulance. There were 3 cars, 2x automated and one human-controlled.
I suspect the human-controlled one did not move in seconds or they wouldn't be talking about delays being a factor in this tragic loss of life. The fact that you didn't pick that up is what makes me think the article is a hit piece.
Sorry, a human driver fatally hit a pedestrian, then a SFPD car was blocking the ambulance from leaving with the injured person? and we're going to make this story about the dangers of Cruise cars? I'm all for complaining if the cruise car did something wrong, but vehicles get disabled in the road all the time for various reasons.
aka the SFPD car was probably there, and cruise cars stalled, blocking the exit so they had to find someone to move one of the 3 cars, 2 being cruise.
The deflection is just weird to me. Can we only discuss issues with self-driving cars once they've replaced human drivers fully?
Pushing a veh is completely appropriate. Especially if its empty. Stalled. Etc.
Done it many times. Can be done by hand or the 30ton truck..
Robot cars, too.